PEAQ在扬声器异常音检测中的应用

扬声器异常音作为产品音质好坏的一种具体表现形式,其检测问题也属于音质评价的一方面。故本文借鉴国际电信联盟推荐的音频质量感知评价算法PEAQ(Perceptual Evaluation of Audio Quality)的实现原理,将其中的人耳听觉模型应用在扬声器的异常音检测中。取标准扬声器样品的声信号作为参考信号,待测扬声器的声信号作为待测信号,两个信号通过人耳听觉模型逐帧处理后,计算得到信号感知差异曲线,用以区分异常音样品。人耳听觉模型包括了中外耳频率加权、听觉滤波器分组、添加内部噪声以及掩蔽效应模拟。将本文所述方法与人工听音以及国际通用的高阶谐波检测扬声器异常音的方法做了比较,实验室检测结果验证了该方法的可行性。

The application of PEAQ to the rub&buzz detection of loudspeaker

As the rub&buzz of loudspeaker is one aspect of audio quality, this paper suggests a new method called PDM to detect the loudspeaker rub&buzz based on the PEAQ algorithm (Perceptual Evaluation of Audio Quality). The differences between the test signals and the reference signal, which are considered by human auditory characteristics, are calculated frame by frame to distinguish tested speakers. The human auditory model contains outer-middle ear frequency weighting, auditory filter groping, internal noise and masking effects in time and frequency domains. Compared with the existing HOD (high order harmonic detection) method, the results of laboratory experiments proved the practicability of the PDM method.